Subdivide group of data

I am having trouble subdiving into sub-groups my data. The first thing I was trying to do is selecting group of data every 30 degrees from 0 to 360. It makes 12 groups of data. The secong thing I would like to do is from this group of data do another selection. For each 30 degrees I would like to divide into 5 groups. Thus have 12 Groups of data and from each group have 5 sub groups. I would like to do, for every 30 degrees residual groups (azimuth), the median of the residuals every 50m (epicenter distance). For example for the group of data with residuals that are between 0-30 degrees I would like to do the median of these residuals with epicentral distance from 0-50, 50-100,..., 250-300. This is the code I tried but it says that the indexe exceeds the matrix, I tried other things but have the same problem:

%dividing for epicenter distance from 0 to 300m every 50m
  N_pieces3=6;

    cheese_angle3=300/N_pieces3;

%diving for azimuth from 0 to 360
  N_pieces=12;
  cheese_angle=360/N_pieces;
%select groups of residuals every 30 degrees and epicentral distance every 30 degrees
    for nps=1:N_pieces
     resp1=resp(azmp1>=cheese_angle*(nps-1) & azmp1<=cheese_angle*nps);
     depip1=depip(azmp1>=cheese_angle*(nps-1) & azmp1<=cheese_angle*nps);

%divind into 6 groups from the previous residual groups depending on the epicentral distance and compute its mean every 50m
        for c=1:N_pieces3
    resp3=resp1(depip>=cheese_angle3*(c-1) & depip<=cheese_angle3*c);
    depip3=depip1(depip>=cheese_angle3*(c-1) & depip<=cheese_angle3*c );
     angle3(c)=cheese_angle3*((c-1)+.5); %for plotting: the point must be at the center
     Med(nps)=median(resp3)-medAs;
        end
   angle2=cheese_angle3*((nps-1)+.5); %for plotting: the point must be in the middle
    end

At the end I would like to do a 3D plotting with the residuals median depending on the azimuth and the epicentral distance. If possible a 3D polar scatter. Thank you.

Answer 1

[Razvan Carbunescu]

If using R2018a you can compute the data easier by using groupsummary.

T = table(cheese,depip,resp);
GT = groupsummary(T,{'cheese','depip'},{12,5},'median','resp');

The above code should group cheese into 12 groups, within those group depip into 5 groups and then compute the median for the data stored in resp for each of the 12*5 groups.

For earlier releases can probably achieve a similar result by using some combination of findgroups / splitapply and discretize.

[Valeria Lorenzoni]

Thank you very much for your help, it is practically to what I was trying to get. The only problem is that I want specifically to take groups of depip that go from 0-50, 50-100, 100-150,150-200. And the problem here is that it divides automatically depip into 5 groups independently to the creteria given (every 50m). Another question, is it possible to plot these values with a 3d polar plotting?

[Razvan Carbunescu]

You can also specify the bins directly if you know they need to be in a certain range.

groupsummary(T,{'cheese','depip'},{0:30:360,0:50:300},'median','resp');

Regarding the plotting aspect I am less familiar but there is a file exchange submission for 3D polar plotting

There are also some previous MATLAB Answers on the topic.

Can probably also use scatter3 after building the x, y coordinates from the groups information.

[Valeria Lorenzoni]

Thank you it worked perfectly!

[Valeria Lorenzoni]

How can I replace the values of each range by its mean. For example for GT(:,1) replace [0, 30) with 15 and so on?

[Razvan Carbunescu]

If you want to do computation afterwards I'd say it's probably easier to convert both to double and use indexing to expand the new names to what you want

newcatscheese = (15:30:345)'; % making into column 
GT.disc_cheese = newcatscheese(double(GT.disc_cheese))

If you want to keep them as categorical can probably change the names of the categories with renamecats.

newcatsdepip = {'25','75','125','175','225','275'};
GT.disc_depip = renamecats(GT.disc_depip,categories(GT.disc_depip),newcatsdepip)